The syntax function[i](){}
implies an object with property values that are functions, function[]
, indexed by the name, [i]
.
Thus
{"f:1":function(){}, "f:2":function(){}, "f:A":function(){}, ... } ["f:"+i]
.
{"f:1":function f1(){}, "f:2":function f2(){}, "f:A":function fA(){}} ["f:"+i]
will preserve function name identification. See notes below regarding :
.
So,
javascript: alert(
new function(a){
this.f={"instance:1":function(){}, "instance:A":function(){}} ["instance:"+a]
}("A") . toSource()
);
displays ({f:(function () {})})
in FireFox.
(This is almost the same idea as this solution, only it uses a generic object and no longer directly populates the window object with the functions.)
This method explicitly populates the environment with instance:x
.
javascript: alert(
new function(a){
this.f=eval("instance:"+a+"="+function(){})
}("A") . toSource()
);
alert(eval("instance:A"));
displays
({f:(function () {})})
and
function () {
}
Though the property function f
references an anonymous function
and not instance:x
, this method avoids several problems with this solution.
javascript: alert(
new function(a){
eval("this.f=function instance"+a+"(){}")
}("A") . toSource()
);
alert(instanceA); /* is undefined outside the object context */
displays only
({f:(function instanceA() {})})
:
makes the javascript function instance:a(){}
invalid.eval
.The following is not necessarily problematic,
instanceA
function is not directly available for use as instanceA()
and so is much more consistent with the original problem context.
Given these considerations,
this.f = {"instance:1": function instance1(){},
"instance:2": function instance2(){},
"instance:A": function instanceA(){},
"instance:Z": function instanceZ(){}
} [ "instance:" + a ]
maintains the global computing environment with the semantics and syntax of the OP example as much as possible.